The hierarchical structural architecture ofinflammasomes, supramolecular inflammatorymachinesArthur V Hauenstein1, Liman Zhang1 and Hao Wu
Available online at www.sciencedirect.com
ScienceDirect
Inflammasomes are caspase-1 activating, molecular
inflammatory machines that proteolytically mature pro-
inflammatory cytokines and induce pyroptotic cell death during
innate immune responses. Recent structural studies of proteins
that constitute inflammasomes have yielded fresh insights into
their assembly mechanisms. In particular, these include a
crystal structure of the CARD-containing NOD-like receptor
NLRC4, the crystallographic and electron microscopy (EM)
studies of the dsDNA sensors AIM2 and IFI16, and of the
regulatory protein p202, and the cryo-EM filament structure of
the PYD domain of the inflammasome adapter ASC. These
data suggest inflammasome assembly that starts with ligand
recognition and release of autoinhibition followed by step-wise
rounds of nucleated polymerization from the sensors to the
adapters, then to caspase-1. In this elegant manner,
inflammasomes form by an ‘all-or-none’ cooperative
mechanism, thereby amplifying the activation of caspase-1.
The dense network of filamentous structures predicted by this
model has been observed in cells as micron-sized puncta.
Address
Department of Biological Chemistry and Molecular Pharmacology,
Harvard Medical School, and Program in Cellular and Molecular
Medicine, Boston Children’s Hospital, Boston, MA 02115, United States
Corresponding author: Wu, Hao ([email protected])
1 These authors contributed equally.
Current Opinion in Structural Biology 2015, 31:75–83
This review comes from a themed issue on Macromolecularmachines and assemblies
Edited by Katrina T Forest and Christopher P Hill
http://dx.doi.org/10.1016/j.sbi.2015.03.014
0959-440X/# 2015 Elsevier Ltd. All rights reserved.
IntroductionThe immune system protects organisms from infections
and other types of insults; It consists of an innate immune
component and an adaptive immune component. Innate
immunity offers the first line of defense and is mediated by
germ line encoded pattern recognition receptors (PRRs)
that recognize pathogen-associated and damage-associated
www.sciencedirect.com
molecular patterns (PAMPs and DAMPs) [1–3]. PRRs
include cell surface and endosomal Toll-like receptors
(TLRs), as well as cytosolic PRRs such as RIG-I-like
receptors (RLRs), AIM2-like receptors (ALRs), NOD-like
receptors (NLRs) and cyclic GMP-AMP synthase (cGAS)
[4�]. PRR signal transduction induces a plethora of cellular
reactions to counter immediate dangers, including cyto-
kine secretion, cell death and interferon response. It also
helps to initiate adaptive immunity for antigen-specific
defense mechanisms.
Over the last decade, the classical model of signal trans-
duction as a series of binding events that trigger the
allosteric activation of enzymes and other downstream
effector molecules has been expanded to include the
formation of large supramolecular assemblies that explain
the threshold kinetics and signal amplification observed in
many innate, as well as adaptive, immune signaling path-
ways [5��,6��,7,8]. These assemblies, which we recently
named supramolecular organizing centers (SMOCs),
often manifest as large, heterogeneous micron-sized
puncta in cells [6��]. Recent advances in cryo-electron
microscopy (cryo-EM) and single-molecule fluorescence
microscopy have begun to bring the structural characteri-
zation of these puncta from the micrometer-scale to the
sub-nanometer-scale [9,10]. In this review, we will focus
on one class of SMOCs, known as inflammasomes, that are
caspase-1 activating machines [11��,12]. The hierarchical
assembly mechanism illustrated here, involving succes-
sive steps of nucleated polymerization, may be general to
other innate immune signaling pathways.
Canonical inflammasomes are formed by the assembly of
three classes of molecules: sensors in the ALR and NLR
family, adapters such as apoptosis-associated, speck-like
protein containing a CARD (ASC) [13], and effectors such
as caspase-1 (Figure 1). Caspase-1 activation leads to
processing of the pro-forms of interleukin-1b (IL-1b)
and IL-18 into their mature forms for secretion, and
induces an inflammatory cell death known as pyroptosis
[11��,12]. ALRs include absent in melanoma 2 (AIM2)
and interferon-inducible protein 16 (IFI16), and are
composed of an N-terminal pyrin domain (PYD) and a
C-terminal, �200 amino acid hematopoietic, interferon-
inducible, nuclear localization (HIN) domain [14–17].
Although the HIN domain detects phagocytosed or
actively replicating viral dsDNA in the cytosol, the
PYD recruits the adapter ASC through homotypic
PYD/PYD interactions. Negative regulators of ALR
Current Opinion in Structural Biology 2015, 31:75–83
76 Macromolecular machines and assemblies
Figure 1
Sensor Proteins
Adapter Proteins
Effector Proteins
AIM2
NLRP3
NAIP5
PYD
PYD NACHT
NACHT
NACHTNLRC4
LRRs
LRRs
LRRs
CARDPYDASC
Caspase-1 CARD P20 P10
BIR
CARD
PYD
HIN200
HIN200 HIN200
HIN-A HIN-BIFI16
p202
AIM2-like receptors (ALRs)
NOD-like receptors (NLRs)
Current Opinion in Structural Biology
Domain structures of inflammasome proteins. Domain abbreviations
are as follows: PYD: Pyrin domain (yellow and red symbols); HIN:
hematopoietic, interferon-inducible, nuclear localization domain (dark
blue rounded rectangles); NACHT: nucleotide-binding and
oligomerization domain (cyan elongated ovals); LRRs: leucine-rich
repeats (repeating orange rectangles); BIR: baculovirus IAP repeat
(dark gray overlapping diamonds); and CARD: caspase recruitment
domain (light green and orange hexagons). Caspase domain: p20 and
p10 as the large and small subunits, respectively (blue and purple
rounded rectangles).
inflammasome formation, such as p202, have only two
HIN domains and lack the PYD [14]. ASC also contains a
CARD domain, which is responsible for caspase-1 recruit-
ment and activation [13]. NLRs have a more complex
domain architecture with variable N-terminal domains,
a central nucleotide binding and oligomerization
(NACHT) domain that shares homology with the
AAA+ superfamily of ATPases, and a C-terminal leucine
rich repeat (LRR) domain [11��,12]. The biggest subfam-
ily of NLRs is the NLRP family, with the ‘P’ representing
the N-terminal PYD domain. The NLRP inflammasomes
require the adapter ASC to mediate caspase-1 activation.
The N-terminal BIR domain-containing NLRs, such as
NAIPs, detect bacterial flagellin and type III secretion
proteins: NAIP5 detects flagellin, and NAIP2 specifically
detects the rod protein of the type III secretion system
Current Opinion in Structural Biology 2015, 31:75–83
[18–20]. Although named as an NLR, the N-terminal
CARD-containing protein NLRC4 is now known to
function as an adapter to the NAIP sensors [18–22]. Upon
ligand binding, NAIPs can interact with NLRC4 to form a
NAIP/NLRC4/caspase-1 inflammasome in the absence
of ASC [18–22]. However, recent studies suggest that
NLRC4 also interacts with ASC, and that the NAIP
inflammasome co-localizes with NLRP3 in a single speck
in THP-1 macrophages upon infection with Salmonellatyphimurium derived flagellin [23].
Inflammasome assembly and caspase activation proceed
in several steps. First, in the resting state, sensor mole-
cules, either ALRs or NLRs, appear to exist in autoin-
hibited conformations. Second, upon encountering
PAMPs and DAMPs such as flagellin, lipoproteins, viral
dsDNA, uric acid crystals and extracellular ATP, ALR or
NLR sensors undergo conformational changes that over-
come the autoinhibition. Third, for sensor proteins with a
PYD such as AIM2 and NLRP3, clustering of PYDs
ensues, which recruits ASC and promotes multivalent
PYD/PYD interactions. Alternatively, for the NAIP
inflammasomes, recruitment of NLRC4 follows, which
leads to clustering of NLRC4 CARDs. Fourth, clustered
ASC and NLRC4 CARDs recruit pro-caspase-1, whose
expression is upregulated by prior NF-kB driven tran-
scription. Activated caspase-1 cleaves pro-IL-1b and
pro-IL-18 to generate the mature forms of these proin-
flammatory cytokines. It may also cleave a collection of
additional substrates to induce the pyroptotic cell death
that is associated with swelling and rupture of cellular
membranes. The following sections will use existing
structural information to illustrate each of these steps
in inflammasome assembly and activation.
Auto-inhibition in ALR and NLR sensor andadapters through intramolecular interactionsBoth inflammasome sensors and adapters have been
proposed to exist in an autoinhibited conformation before
activation. A structure of the mouse NLRC4 adapter
lacking the N-terminal CARD (DCARD mNLRC4) pro-
vides insights about the mechanism of autoinhibition in
this NLR. NLRC4 functions as an adapter protein in
NAIP/NLRC4/caspase-1 inflammasome formation [24��].The NACHT domain of NLRC4 is composed of a
nucleotide-binding domain (NBD), a helical domain 1
(HD1), a winged helix domain (WHD), and a helical
domain 2 (HD2) (Figure 2a). The autoinhibited
mNLRC4 assumes a closed conformation with intricate
intramolecular interactions that sequester mNLRC4 in a
monomeric state (Figure 2a). Without oligomerization,
the CARD of NLRC4 is unable to recruit caspase-1
through CARD–CARD interactions, thereby affording
a safety mechanism against auto-activation in the absence
of proper stimulation. In the autoinhibited DCARD
mNLRC4 structure, an ADP molecule is bound at the
interface between the NBD and WHD (Figure 2a). The
www.sciencedirect.com
The hierarchical structure of inflammasomes Hauenstein, Zhang and Wu 77
Figure 2
NBDHD1
WHD
HD2
LRR
(a)
CED4 NBDNLRC4 NBD
AIM2
ADP
S533p
NBD oligomerization interface
(b)
(c)
NBD HD1 WHD HD2 LRRCARD
CARD
PYD HIN
NACHT
Current Opinion in Structural Biology
Structure of monomeric, inactive DCARD mouse NLRC4. (a) The NACHT domain of NLRC4 is divided as follows: NBD: nucleotide-binding domain
(cyan); HD1: helical domain 1 (blue); WHD: winged helix domain (yellow); and HD2: helical domain 2 (magenta). LRR: leucine-rich repeat (orange).
(b) Structural superposition of the inactive NLRC4 NBD (cyan) with two neighboring NBDs (gray) of active CED-4 in the C. elegans caspase
activating complex apoptosome (PDB ID: 3LQQ) [25]. The inactive NLRC4 NBD is in a conformation incompatible with NBD oligomerization. (c)
Schematic model shows autoinhibition of AIM2 through intramolecular interactions between the PYD and the HIN200 domain.
HD2 interacts with a functionally important site in NBD
and caps the N-terminal side of LRR through a phos-
phorylated serine residue S533p. The C-terminal LRR
domain directly binds one side of the NBD to sterically
prevent it from self-oligomerization. Maintenance of the
autoinhibited state must be cooperative as disruption of
any of the above mentioned interactions cause constitu-
tive activation of mNLRC4 and caspase-1 [24��]. In
addition, the mNLRC4 NBD is in a conformation that
is incompatible with oligomerization as shown by its
superposition with two adjacent NBDs in the octameric
CED-4 [25]. Several helices of the NBD would clash with
a neighboring NBD molecule in the octameric assembly
(Figure 2b). Presumably, other NLRs, whether they are
sensors or adapters, may use a similar structural mecha-
nism for their autoinhibition.
Although the autoinhibition mechanism of ALR sensors
such as AIM2 has not been directly visualized through
structural studies, it has been proposed that the
N-terminal PYD and the C-terminal HIN domains form
intramolecular interactions to prevent PYD oligomeri-
zation (Figure 2c). For AIM2, its PYD itself is able to
form filaments and therefore promotes inflammasome
assembly and activation in the absence of the HIN
domain [26�,27��]. The direct interaction between
www.sciencedirect.com
AIM2 PYD and its HIN domain has been measured
and confirmed [28�], consistent with a mechanism of
autoinhibition through intramolecular interactions. A
docking model shows that AIM2 PYD may bind AIM2
HIN at the same surface that the HIN domain binds
dsDNA; thus DNA binding may release autoinhibited
PYD from the HIN domain, thereby facilitating AIM2
PYD oligomerization and activating ASC through
PYD/PYD interactions [28�,29��]. There is a long,
�50-residue linker between the PYD and the HIN
domain, which should allow structural rearrangement
upon activation.
DNA recognition and regulation by ALR HIN domains
Ligand binding may be a general mechanism to overcome
autoinhibition, leading to initiation of inflammasome
assembly. Although it is presumed that the C-terminal
domain of an NLR is the receptor for stimulation and acts
to relieve autoinhibition, the molecular basis for this
function is unclear. By contrast, for ALRs, the mode of
direct sensing of cytosolic dsDNA by their HIN domains
has been revealed from a series of structural studies. Upon
binding dsDNA, ALRs apparently undergo a conforma-
tional change to disengage from their autoinhibited states
to allow the freed N-terminal PYD to cluster and initiate
inflammasome activation [26�,28�,29��].
Current Opinion in Structural Biology 2015, 31:75–83
78 Macromolecular machines and assemblies
The HIN domain is comprised of tandem oligonucleo-
tide/oligosaccharide binding (OB) folds (Figure 3a). Crys-
tal structures of dsDNA-bound HIN domains from AIM2
and IFI16 show a similar mode of interaction that involves
both OB folds and the intervening linker [29��,30�].Highly positively charged surfaces of AIM2 HIN and
IFI16 HIN-B domains contact the dsDNA via mostly
electrostatic interactions (Figure 3a,b). By contrast to
sequence-specific DNA-binding, in which the proteins
directly recognize the DNA bases, the HIN domains
interact almost exclusively with the backbone. AIM2
and IFI16 HIN domains bind both strands of dsDNA,
explaining why dsDNA, but not ssDNA, is able to initiate
innate immune responses through AIM2 and IFI16.
Interactions of ALRs with long dsDNA appear to deviate
from the presumed random distribution of binding im-
plied by a beads on a string model. Instead, long dsDNA
appears to cause cooperative binding of ALRs. For exam-
ple, full-length IFI16 has been shown to cooperatively
bind dsDNA in a length-dependent manner into distinct
IFI16/dsDNA filaments even in the presence of excess
dsDNA [31�] (Figure 3c), suggesting the participation
of protein/protein interactions. Although there are
HIN/HIN contacts in the crystal structures of HIN/
dsDNA complex structures [29��,30�], the PYD of
IFI16 is crucial for the formation of IFI16 filaments
[31�]. These data indicate that intermolecular interac-
tions among the PYDs drive the cooperative assembly
onto dsDNA, with concomitant increase in the IFI16/
dsDNA binding affinity when compared with isolated
HIN domains alone [31�]. This suggests that the
dsDNA filaments and PYD filaments may form concur-
rently, leading to a central dsDNA filament decorated
by short PYD filaments along its length [26�,28�](Figure 3c).
ALR inflammasome activation is tightly regulated in
cells. An example comes from the mouse p202 protein,
which lacks the N-terminal PYD but contains two HIN
domains [14]. Expression of p202 is induced by interfer-
on, and as negative feedback regulation p202 inhibits the
inflammatory function of AIM2 [14]. Crystal structures of
the p202 HIN1/dsDNA complexes show that the two OB
folds are very similar to those of AIM2, but that the
dsDNA binds p202 HIN1 at an almost opposite surface
compared with that used in AIM2 HIN (Figure 3d,e)
[30�,32,33��]. Surprisingly, the HIN2 domain of p202
does not interact with dsDNA [33��]. It forms a dimer-
of-dimers tetramer, utilizing the equivalent surface in
HIN1 for dsDNA interaction as one of the dimerization
interfaces (Figure 3e). The HIN2 of p202 also interacts
specifically with the AIM2 HIN domain [33��]. There-
fore, the inhibitory function of p202 to AIM2 may be
facilitated through two mechanisms. First, p202 binds
dsDNA at a higher affinity than AIM2, and therefore
competes with AIM2 for access to dsDNA. Second, the
Current Opinion in Structural Biology 2015, 31:75–83
direct interaction between p202 HIN2 and AIM2 HIN
domains may allow incorporation of p202 onto the same
dsDNA that AIM2 interacts with, thereby diluting the
effective concentration of AIM2 on the dsDNA
[30�,32,33��] (Figure 3f). Thus, AIM2 PYD oligomeriza-
tion and downstream signaling may be inhibited.
The nucleated polymerization mechanism of
inflammasome assembly
The PYDs and CARDs involved in inflammasome
assembly both belong to the death domain superfamily
that mediates the formation of oligomeric signaling com-
plexes important in cell death and innate immune signal-
ing [34��,35,36,37�]. The subunit structures of these
domains exhibit a globular, six-helix bundle fold, as
illustrated here for ASC PYD [27��] (Figure 4a). For
ASC-dependent inflammasomes, the central organizing
scaffold is formed by PYD/PYD and CARD/CARD inter-
actions through a common helical assembly mechanism
that was first elucidated for death domain complexes such
as the oligomeric MyD88/IRAK4/IRAK2 complex and
the PIDD/RAIDD complex [34��,37�].
Because of the importance of PYDs in inflammasome
assembly, many structural studies have been performed
on these domains. However, like many death domain
superfamily proteins, PYDs are multivalent with a high
propensity to aggregate, which makes structural studies
difficult. To reduce aggregation for NMR and X-ray
crystallography, many groups have prepared PYD pro-
teins at low pH (�pH 4), which likely modifies the surface
potential and prevents charge complementarity-induced
aggregation. To date twelve monomeric PYD domain
structures have been solved including ASC [38�,39],
AIM2 [26�,28�], NLRP1 [40], NLRP3 [41], NLRP4
[42], NLRP7 [43], NLRP10 (human) [44], NLRP10
(mouse, PDB ID: 2DO9), NLRP12 [45], ASC2 (a regula-
tory PYD only protein) [46,47], myeloid nuclear differen-
tiation antigen (MNDA, PDB ID: 2DBG), and Pyrin
(PDB ID: 2MPC, mutations of which cause familial
Mediterranean fever).
Recent studies show that ASC PYD forms filamentous
structures [27��,48�], and that AIM2 PYD and NLRP3
can both form a complex with ASC PYD [27��]. Electron
microscopy imaging shows that the AIM2-PYD/ASC-
PYD and NLRP3/ASC-PYD complexes are also filamen-
tous [27��]. Gold-particle labeling of AIM2 PYD and
NLRP3 in these complexes reveal that they only reside
at one end of the filamentous structures that are com-
posed mostly of ASC PYD (Figure 4b,c), suggesting that
the inflammasome sensors nucleate filament formation of
the ASC adapter [27��]. Indeed, fluorescence polarization
assays reveal that sub-stoichiometric amounts of the
AIM2/dsDNA complex or oligomerized NLRP3 can
robustly nucleate the formation of the ASC-PYD filament
[27��].
www.sciencedirect.com
The hierarchical structure of inflammasomes Hauenstein, Zhang and Wu 79
Figure 3
PYDAIM2
IFI1
6/d
sDN
A(6
00b
p)
com
ple
xHIN
OB1 PYDIFI16 OB2HIN-B
OB1HIN-AOB2
–90°–90°
P202 OB2HIN2
OB1OB2OB1HIN1
90°
OB1
OB2
OB1
OB2
OB1
OB2
PYD
AIM2 HIN/dsDN A
100 nm
100 nm
IFI16 HIN-B/dsDNA
p202 HIN2/dsDNA
AIM2 HIN/dsDNA
p202 HIN1/dsDNA
p202 HIN2tetramer
OB1
OB2
dsDNA
HIN-A HIN-B
(a) (b)
(c) (d)
(e) (f)
PYD
AIM2 PYD p202 HIN1AIM2 HIN p202 HIN2
p202
AIM2
AIM2
dsDNA
dsDNA
Clustering
No Clustering+
Current Opinion in Structural Biology
Ligand binding, filament assembly and negative regulation of ALRs. (a) Left: structure of AIM2 PYD (yellow). Middle and right: the HIN/dsDNA
complex structure shown respectively in a ribbon diagram and in a surface electrostatics illustration rotated 908 along the vertical axis. HIN: dark
blue. (b) Structure of the IFI16 HIN-B/dsDNA complex shown respectively in a ribbon diagram and in a surface electrostatics illustration rotated
�908 along the vertical axis. (c) EM images (top, adapted from Figure 5 of [31�]) and a schematic model (bottom) of filamentous structures of the
IFI16/dsDNA complex. (d) Structure of the p202 HIN1/dsDNA complex shown respectively in a ribbon diagram and in a surface electrostatics
illustration rotated �908 along the vertical axis. (e) Left: p202 HIN2 tetramer structure in a ribbon diagram with monomers shown in red, yellow,
cyan and pink. Right: the same structure in a surface diagram, with the red monomer superimposed with the p202 HIN1/dsDNA structure (orange)
and the AIM2 HIN/dsDNA structure (gray). One HIN2 tetramerization surface is analogous to the p202 HIN1 surface for dsDNA recognition, but
opposite of the AIM2 HIN surface for dsDNA recognition. (f) Schematic mechanistic models for the interference with AIM2 signaling by p202
(adapted from Figure 6 of [33��]).
www.sciencedirect.com Current Opinion in Structural Biology 2015, 31:75–83
80 Macromolecular machines and assemblies
Figure 4
α2- α3 loop
ASC PYD monomer
(a)
(c)
(e)
(f)
(d)
(b)
AIM2 PYD cluster
AIM2 PYD/ASC PYD complex
AIM2 PYD/ASC FL/Caspase-1 CARD complex
Streptavidin-gold labeling ofBiotin-AIMPYD/ASCPYD filaments
Ni-NTA gold (6 nm) labelingof His-GFP-caspase-1 CARD
Inner diameter
Outer diameter
ASC PYD filament
Anti-ASC gold (15 nm)
28 nm
Flagellin/NAIP5/NLRC4 complex
AIM2 PYD/ASC FL complex
Top view Side view
100nm
100nm
AS
C C
AR
D
20 Å
90º
90 Å
90º
α5α6
α1α2
α3 α4
Current Opinion in Structural Biology
ASC PYD filament structure and mechanism of nucleation dependent polymerization in Inflammasome formation. (a) Near atomic resolution cryo-
EM structure of a protomer from the ASC PYD filament (PDB ID: 3J63). (b) ASC PYD helical filament structure is depicted as a surface
representation in two orthogonal views, and is composed of three helical strands denoted by the colors green, blue, and red. (c) The AIM2 PYD
structure (PDB ID: 3VD8) is modeled as a cluster with C3 symmetry (depicted as protomers in shades of yellow to denote subunit boundaries) at
the end of the ASC PYD filament. An electron micrograph with streptavidin-gold labeling of biotin-AIM2 PYD/ASC PYD filaments shows that AIM2
PYD localizes to one end of the ASC PYD filament (adapted from Figure 1D of [27��]). (d) The full-length ASC solution structure (PDB ID: 2KN6) is
modeled as a cartoon-ribbon representation onto the ASC PYD filament structure (transparent surface representation) showing the outward
projecting ASC-CARD domains. The AIM2 PYD cluster is also modeled in shades of yellow. Two orthogonal views are shown. (e) Electron
micrographs of the His-GFP-caspase-1 CARD/ASC/AIM2 PYD ternary complex with anti-ASC gold and Ni-NTA gold labeling respectively, adapted
from Figure 6 of [27��]. (f) Electron micrographs of the flagellin/NAIP5/NLRC4 complex purified from HEK293E cells adapted from Figure 8 of [50].
The cryo-EM structure of the ASC-PYD filament pro-
vides a first view of the inflammasome assembly mecha-
nism at near atomic resolution [27��]. The filament
resembles a cylinder with outer and inner diameters of
Current Opinion in Structural Biology 2015, 31:75–83
�90 A and �20 A, respectively, and is built from three
helical strands related by a three-fold symmetry along the
helical axis (Figure 4b). Three types of repeating asym-
metric interfaces that are characteristic of death domain
www.sciencedirect.com
The hierarchical structure of inflammasomes Hauenstein, Zhang and Wu 81
superfamily proteins (Type I, II, and III) are used to form
the filament from individual ASC-PYD protomers [34��].Upon incorporation into the filament, significant confor-
mational changes occur in ASC-PYD, especially at the
a2–a3 loop and the a3 helix, suggesting a directional
elongation in ASC-PYD filament formation [27��].
Because AIM2-PYD also forms filaments when expressed
alone [26�], it is anticipated that the filament will match
the symmetry of ASC-PYD and therefore is able to
nucleate ASC-PYD filament formation upon activation
by cytosolic dsDNA. A model of the AIM2-PYD filament
can be obtained by superimposing the AIM2-PYD pro-
tomer structure onto the ASC-PYD filament (Figure 3c).
Connecting the short AIM2-PYD filament that is likely to
form when wrapped around dsDNA (Figure 3c) and the
long ASC-PYD filament would recapitulate the end-la-
beled AIM2/ASC complex observed by electron micros-
copy (Figure 4c), thereby providing the molecular basis
for the nucleation and polymerization relationship be-
tween AIM2 and ASC.
In addition to the N-terminal PYD, ASC is a bifunctional
molecule with a C-terminal CARD. The NMR structure of
full-length ASC determined at a non-aggregating condition
shows that the two domains are flexibly linked but with
some preference in the relative orientation [38�]. Super-
imposing the full-length ASC structure onto the ASC-PYD
filament structure generates a model of the full-length ASC
filament in which the PYD localizes at the core and the
CARD forms the outer layer of the filament (Figure 4d).
Radially positioned CARDs can then further cluster,
recruit and nucleate filament formation of caspase-1, which
is upregulated from previous NF-kB driven transcription,
through CARD/CARD interactions of currently unknown
Figure 5
AIM2
DNA
ATP, K+ efflux
PYDPYD
LRHIN
200
NACHT
AIM2
NLRP3NLRP3 PYDClustering
AIM2 PYDClustering
PYD-PYD InASC Filame
A schematic diagram for the nucleation dependent polymerization mechanis
induced autoproteolytic maturation of caspase-1. Proteins and domains are
www.sciencedirect.com
mechanisms. The reconstituted ternary AIM2-PYD/ASC/
caspase-1-CARD complex shows star-shaped structures
under electron microscopy in which AIM2 and ASC locate
near the center while caspase-1 forms the main bodies of
the filamentous extensions, as shown by specific gold-
labeling [27��] (Figure 4e). The ASC-dependent NLRP3
inflammasome assembles through a similar mechanism.
The initial star-shaped structures likely further coalesce
to become a dense single perinuclear punctum per cell
composed of networks of ASC and casapse-1 filaments
[15,27��,49].
The mechanism of ASC-independent NAIP/NLRC4/
caspase-1 inflammasome assembly is still unclear as there
are no atomic resolution structures currently available. A
preliminary electron microscopy study shows that the
Flagellin/NAIP5/NLRC4 complex forms a striking dou-
ble disk structure with 11-fold or 12-fold symmetry [50]
(Figure 4f). Presumably, the CARDs of NLRC4 in the
double disk structure are clustered to allow the recruit-
ment of caspase-1 and nucleation of caspase-1 filament
formation, leading to proximity-induced, auto-proteolytic
cleavage of caspase-1 to yield its active form. Because
ASC appears to enhance NAIP inflammasome formation,
it is likely that the NLRC4 CARD can interact with the
ASC CARD, which in turn recruits more caspase-1
through multiple CARD/CARD interactions to amplify
the activation signal.
Conclusions and perspectivesRecent structural studies have now provided a framework
for understanding inflammasomes through steps of hier-
archical assembly involving ligand recognition, release of
autoinhibition, and nucleated polymerization (Figure 5).
The process is likely highly cooperative, generating an
PYDRsCARD
CARDP20 P10
Pro-Caspase-1
Caspase-1
teractionsnt Growth
NLRP3 ASC
CARD-CARD InteractionsPro-Caspase-1 Filament Growth
Current Opinion in Structural Biology
m of AIM2 and NLRP3 inflammasome formation, with proximity-
labeled.
Current Opinion in Structural Biology 2015, 31:75–83
82 Macromolecular machines and assemblies
all-or-none mechanism of inflammasome activation,
which is a common emerging theme among SMOCs in
innate immunity [5��,6��]. The increase in stoichiometry
from the ALR and NLR sensors, to the adapter ASC, and
then to caspase-1 in the ternary inflammasomes, gives an
effective amplification mechanism for the signal trans-
duction. For the NLRP3 inflammasome, the efficient
assembly and amplification is reflected in their low acti-
vation threshold in dendritic cells [51]. Although we have
progressed a great deal, many more structural studies are
required to elucidate the detailed molecular mechanism
for the assembly and activation of each member of the
large inflammasome family, undoubtedly with surprises
awaiting.
Conflict of interest statementNone declared.
Acknowledgments
We apologize for incomplete coverage due to the space limitations and thevast data in the highly active field of inflammasome biology. The work wassupported by grants AI050872, AI045937, and AI089882 from NIH to HW.
References and recommended readingPapers of particular interest, published within the period of review,have been highlighted as:
� of special interest�� of outstanding interest
1. Janeway CA Jr: Approaching the asymptote? Evolution andrevolution in immunology. Cold Spring Harb Symp Quant Biol1989, 54(Pt 1):1-13.
2. Poltorak A et al.: Defective LPS signaling in C3H/HeJ andC57BL/10ScCr mice: mutations in Tlr4 gene. Science 1998,282:2085-2088.
3. Medzhitov R et al.: A human homologue of the Drosophila Tollprotein signals activation of adaptive immunity. Nature 1997,388:394-397.
4.�
Yin Q et al.: Structural biology of innate immunity. Ann RevImmunol 2015, 33:13.1-13.24.
This review gives a structural overview of proteins that constitute varioussupramolecular signaling complexes in the innate immune response.
5.��
Wu H: Higher-order assemblies in a new paradigm of signaltransduction. Cell 2013, 153:287-292.
This paper describes the growing viewpoint that innate immune signalingis dominated by the cooperative assembly of large, multimeric proteincomplexes, which drive proximity-driven enzyme activation instead of theclassical model of signal transduction as a cascade of protein–proteinbinding and enzyme activation events.
6.��
Kagan JC et al.: Supramolecular organizing centres: location-specific higher-order signalling complexes that control innateimmunity. Nat Rev Immunol 2014, 14:821-826.
This review coins the term, supramolecular organizing centers (SMOCs),as a way to describe spatially resolved higher-order immune signalingcomplexes.
7. Sherman E et al.: Super-resolution characterization of TCR-dependent signaling clusters. Immunol Rev 2013, 251:21-35.
8. Paul S, Schaefer BC: A new look at T cell receptor signaling tonuclear factor-kappaB. Trends Immunol 2013, 34:269-281.
9. Kuhlbrandt W: Cryo-EM enters a new era. eLife 2014, 3:e03678.
10. Godin AG et al.: Super-resolution microscopy approaches forlive cell imaging. Biophys J 2014, 107:1777-1784.
Current Opinion in Structural Biology 2015, 31:75–83
11.��
Lamkanfi M, Dixit VM: Mechanisms and functions ofinflammasomes. Cell 2014, 157:1013-1022.
This review describes canonical and non-canonical inflammasomebiology with overviews of NLR and ALR inflammasome assembly andactivation, and of implications on therapeutic applications.
12. Rathinam VA et al.: Regulation of inflammasome signaling.Nat Immunol 2012, 13 333-332.
13. Masumoto J et al.: ASC, a novel 22-kDa protein, aggregatesduring apoptosis of human promyelocytic leukemia HL-60cells. J Biol Chem 1999, 274:33835-33838.
14. Roberts TL et al.: HIN-200 proteins regulate caspase activationin response to foreign cytoplasmic DNA. Science 2009,323:1057-1060.
15. Hornung V et al.: AIM2 recognizes cytosolic dsDNA and forms acaspase-1-activating inflammasome with ASC. Nature 2009,458:514-518.
16. Fernandes-Alnemri T et al.: AIM2 activates the inflammasomeand cell death in response to cytoplasmic DNA. Nature 2009,458:509-513.
17. Burckstummer T et al.: An orthogonal proteomic-genomicscreen identifies AIM2 as a cytoplasmic DNA sensor for theinflammasome. Nat Immunol 2009, 10:266-272.
18. Kofoed EM, Vance RE: Innate immune recognition of bacterialligands by NAIPs determines inflammasome specificity.Nature 2011, 477:592-595.
19. Zhao Y et al.: The NLRC4 inflammasome receptors for bacterialflagellin and type III secretion apparatus. Nature 2011, 477:596-600.
20. Yang J et al.: Human NAIP and mouse NAIP1 recognizebacterial type III secretion needle protein for inflammasomeactivation. Proc Natl Acad Sci U S A 2013, 110:14408-14413.
21. Poyet JL et al.: Identification of Ipaf, a human caspase-1-activating protein related to Apaf-1. J Biol Chem 2001,276:28309-28313.
22. Mariathasan S et al.: Differential activation of theinflammasome by caspase-1 adaptors ASC and Ipaf. Nature2004, 430:213-218.
23. Man SM et al.: Inflammasome activation causes dualrecruitment of NLRC4 and NLRP3 to the samemacromolecular complex. Proc Natl Acad Sci U S A 2014,111:7403-7408.
24.��
Hu Z et al.: Crystal structure of NLRC4 reveals its autoinhibitionmechanism. Science 2013, 341:172-175.
This paper reports the first crystal structure of a CARD-containing NOD-like receptor. The observed closed conformation mediated by extensiveintramolecular interactions suggests an autoinhibition mechanism thatprevents oligomerization.
25. Qi S et al.: Crystal structure of the Caenorhabditis elegansapoptosome reveals an octameric assembly of CED-4. Cell2010, 141:446-457.
26.�
Lu A et al.: Crystal structure of the F27G AIM2 PYD mutantand similarities of its self-association to DED/DEDinteractions. J Mol Biol 2014, 426:1420-1427.
This study describes the filament forming ability of AIM2-PYD and thecrystal structure of the monomeric F27G AIM2-PYD mutant.
27.��
Lu A et al.: Unified polymerization mechanism for the assemblyof ASC-dependent inflammasomes. Cell 2014, 156:1193-1206.
This paper reports the first near atomic resolution cryo-EM structure ofthe ASC-PYD filament, and in vitro reconstitution of the PYD/PYD andCARD/CARD interactions in the AIM2 and NLRP3 inflammasome. Thesestudies reveal a step-wise nucleated polymerization mechanism ofinflammasome assembly.
28.�
Jin T et al.: Structure of the absent in melanoma 2 (AIM2) pyrindomain provides insights into the mechanisms of AIM2autoinhibition and inflammasome assembly. J Biol Chem 2013,288:13225-13235.
This article presents the MBP-fused AIM2-PYD crystal structure andprovides evidence for an intramolecular interaction between the PYDand the HIN domain of AIM2 in autoinhibition.
www.sciencedirect.com
The hierarchical structure of inflammasomes Hauenstein, Zhang and Wu 83
29.��
Jin T et al.: Structures of the HIN domain:DNA complexesreveal ligand binding and activation mechanisms of the AIM2inflammasome and IFI16 receptor. Immunity 2012, 36:561-571.
The crystal structures of AIM2-HIN200 and IFI16-HIN-B each bound todsDNA are presented. HIN binding to dsDNA is shown to be non-specificand dependent on charge/charge interactions.
30.�
Ru H et al.: Structural basis for termination of AIM2-mediatedsignaling by p202. Cell Res 2013, 23:855-858.
This paper presents crystal structures of AIM2-HIN200/dsDNA and p202-HIN1/dsDNA complexes. It further describes the molecular basis for theregulatory function of p202 by comparing the DNA binding modes ofAIM2-HIN200 versus p202-HIN1.
31.�
Morrone SR et al.: Cooperative assembly of IFI16 filaments ondsDNA provides insights into host defense strategy. Proc NatlAcad Sci U S A 2014, 111 E62–71.
This report presents EM and quantitative binding data, which suggest thatIFI16 binds dsDNA in a cooperative manner and that cooperative bindingof the HIN domains to dsDNA is driven by the PYD.
32. Li H et al.: Structural mechanism of DNA recognition by thep202 HINa domain: insights into the inhibition of Aim2-mediated inflammatory signalling. Acta Crystallogr Sect F StructBiol Commun 2014, 70:21-29.
33.��
Yin Q et al.: Molecular mechanism for p202-mediated specificinhibition of AIM2 inflammasome activation. Cell Rep 2013,4:327-339.
Crystallographic and EM structures of tetrameric p202 and the p202-HIN1 domain bound to dsDNA are presented. These structures, alongwith quantitative binding data showing a direct interaction between p202-HIN2 and AIM2-HIN200 domains, suggest an inhibitory mechanism inwhich p202 tetramers bind both dsDNA and AIM2, preventing coopera-tive clustering of AIM2-PYDs.
34.��
Ferrao R, Wu H: Helical assembly in the death domain (DD)superfamily. Curr Opin Struct Biol 2012, 22:241-247.
This review describes the conserved three types of interactions thatmediate the helical assembly of death domain containing proteins includ-ing the 5 PIDD/7 RAIDD, 6 MyD88/4 IRAK4/4 IRAK2, and 5 Fas/5 FADDsignaling complexes.
35. Park HH et al.: Death domain assembly mechanism revealed bycrystal structure of the oligomeric PIDDosome core complex.Cell 2007, 128:533-546.
36. Wang L et al.: The Fas-FADD death domain complex structurereveals the basis of DISC assembly and disease mutations. NatStruct Mol Biol 2010, 17:1324-1329.
37.�
Lin SC et al.: Helical assembly in the MyD88–IRAK4–IRAK2complex in TLR/IL-1R signalling. Nature 2010, 465:885-890.
The crystal structure of the 14-subunit MyD88/IRAK4/IRAK2 deathdomain complex is reported, revealing a helical assembly mechanism.
38.�
de Alba E: Structure and interdomain dynamics of apoptosis-associated speck-like protein containing a CARD (ASC). J BiolChem 2009, 284:32932-32941.
www.sciencedirect.com
This article presents the solution structure of full-length ASC. Despite along flexible linker connecting the PYD and CARD domains, a preferredback-to-back domain orientation is observed.
39. Liepinsh E et al.: The death-domain fold of the ASC PYRINdomain, presenting a basis for PYRIN/PYRIN recognition.J Mol Biol 2003, 332:1155-1163.
40. Hiller S et al.: NMR structure of the apoptosis- andinflammation-related NALP1 pyrin domain. Structure (Camb)2003, 11:1199-1205.
41. Bae JY, Park HH: Crystal structure of NALP3 protein pyrindomain (PYD) and its implications in inflammasome assembly.J Biol Chem 2011, 286:39528-39536.
42. Eibl C et al.: Structural and functional analysis of the NLRP4pyrin domain. Biochemistry 2012, 51:7330-7341.
43. Pinheiro AS et al.: Three-dimensional structure of the NLRP7pyrin domain: insight into pyrin–pyrin-mediated effectordomain signaling in innate immunity. J Biol Chem 2010,285:27402-27410.
44. Su MY et al.: Three-dimensional structure of humanNLRP10/PYNOD pyrin domain reveals a homotypic interactionsite distinct from its mouse homologue. PLoS One 2013,8:e67843.
45. Pinheiro AS et al.: The NLRP12 pyrin domain: structure,dynamics, and functional insights. J Mol Biol 2011,413:790-803.
46. Natarajan A et al.: Structure and dynamics of ASC2, a pyrindomain-only protein that regulates inflammatory signaling.J Biol Chem 2006, 281:31863-31875.
47. Espejo F, Patarroyo ME: Determining the 3D structure of humanASC2 protein involved in apoptosis and inflammation.Biochem Biophys Res Commun 2006, 340:860-864.
48.�
Cai X et al.: Prion-like polymerization underlies signaltransduction in antiviral immune defense and inflammasomeactivation. Cell 2014, 156:1207-1222.
Presents data showing PYD and CARD containing proteins such as ASCand MAVS have inducible prion-like behavior in yeast and mammalian cellreconstitution assays.
49. Wu J et al.: Involvement of the AIM2, NLRC4, and NLRP3inflammasomes in caspase-1 activation by Listeriamonocytogenes. J Clin Immunol 2010, 30:693-702.
50. Halff EF et al.: Formation and structure of a NAIP5-NLRC4inflammasome induced by direct interactions with conservedN- and C-terminal regions of flagellin. J Biol Chem 2012,287:38460-38472.
51. Latz E et al.: Activation and regulation of the inflammasomes.Nat Rev Immunol 2013, 13:397-411.
Current Opinion in Structural Biology 2015, 31:75–83